台灣積體電路製造產業之效率分析

台灣晶圓代工產業位居全球領先地位，2017年代工產值為全球第一，然而面臨中國積極擴張積體電路產業投資基金規模，且高階應用市場的需求崛起下，半導體產業維持甚至提升競爭優勢顯得尤為重要。運用資料包絡分析法可以找出廠商間相對效率的分布，並找出標竿廠商，作為效率改進之目標。本文以面臨相類似的產業環境的台灣12家積體電路製造廠商作為分析對象，以CCR與BCC模式分析2018年各廠的效率表現，並以麥氏生產力指數作為分析2016~2018年效率變動之指標。本研究分析整體積體電路製造產業與個別廠商指標無效率之原因。並進一步以靜態與動態DEA指標評析台灣IC製造廠商的績效，以免過度聚焦於研究時點的績效表現而忽略多期的績效改進。得出在研究時點2018年大多廠商的總技術效率都處於中等效率，尚具改善空間，然而從跨期變動指標則看出大多數廠商總技術效率呈現改善的趨勢。本研究並根據以上指標，繪製以動靜態為指標的績效評估矩陣，分析總要素生產力變動分別與靜態指標（2018年的效率表現）及動態指標（2016~2018年效率變動）之關係，並探討廠商之間相對的績效表現，並從中發現效率指標領先的廠商並作他廠學習之標竿。

關鍵字

資料包絡分析法；靜態指標；動態指標；績效評估矩陣；台灣積體電路製造廠商

並列摘要

Taiwanese Semiconductor Foundry industry has the highest production value over the world in 2017. However, China is making a huge investment over its semiconductor industry as a part of Made in China in 2025. In addition, under the growing demand of new application markets, to keep the competitive advantage is significantly important for the whole semiconductor industry and also for Taiwanese Integrated circuit (abbreviated below as IC) manufacturing firms. By applying Data Envelope Analysis Method (DEA) could find the relative performance of firms as well as finding the benchmark which could be regarded as a goal of efficiency improvement. The twelve IC manufacturing firms which are facing similar business environment are selected in this study. The CCR and BCC model would be applied to analyze the static efficiency of 2018. The Malmquist Productivity Index would be applied to analyze the dynamic efficiency change from 2016 to 2018. The reason for the inefficiency of the whole industry and separate firms would be looked into in this study. Further, the performance would be evaluated from both static and dynamic efficiency index in order not to just focus on a single period performance but ignore the cross-section efficiency improvement. This research finds that almost all the firms are with a middle-efficiency level in 2018, so they still require efficiency improvement. But when it comes to dynamic efficiency index, almost all the firms are improving their total factor productivity. This study plotted the performance evaluation matrix based on the static (the efficiency score of 2018) and dynamic index (the efficiency change during 2016 to 2018) to find relative efficiency of firms, and also find the benchmark that could be learned from other firms.